Composite Part B, 46:211–220, 2013.

Determination of Material Parameters for

Abaqus Progressive Damage Analysis of
E-Glass Epoxy Laminates

E. J. Barbero1, F. A. Cosso 2, R. Roman2, T. L. Weadon2

Mechanical and Aerospace Engineering, West Virginia University,
Morgantown, WV 26506-6106, USA

abstract
A methodology for determination of material parameters for the progressive damage analysis (PDA) model
implemented in Abaqus is presented. The methodology is based on fitting PDA model results to available
experimental data. The applicability of the PDA model is studied by comparison to a broad set of ex-
perimental data for E-Glass Epoxy laminates, as well as contrasting PDA with ply discount results.Also,
sensitivity of the Abaqus PDA model to h- and p-refinement is studied.

keyword
A. Polymer-matrix composites (PMCs); B. Transverse cracking; C. Damage mechanics; C. Finite element
analysis (FEA); Parameter Identification.

1 Introduction
Prediction of damage initiation and accumulation in polymer matrix, laminated composites is of great interest
for the design, production, certification, and monitoring of an increasingly large variety of structures. It is
also a complex and difficult topic. Only matrix cracking due to transverse tensile and shear deformations is
considered in this manuscript. Matrix cracking is normally the first mode of damage. If left unmitigated,
it often leads to other modes such as delamination and, due to load redistribution, may even lead to fiber
failure of adjacent laminas.
The earliest, simplest and least accurate modeling technique to address matrix damage is perhaps the
ply discount method [1, Section 7.3.1]. Ply discount is used here as a baseline for contrasting predictions
obtained with the progressive damage analysis (PDA) method implemented in AbaqusTM3 [?]. Although
many other models exist [2–12], etc., and several plugins are available [13, 14], this manuscript focuses on
the PDA model available in Abaqus. This is because of the broad availability of Abaqus, while plugins are
not free, and the remaining models available in the literature are for the most part not readily available to
be used in conjunction with commercial FEA environments.
Therefore, the objective of this manuscript is to propose a methodology to determine values for the
material parameters required by the Abaqus PDA model. In this work, the values for the parameters are
found using available experimental data and a rational procedure. Once values are found, the PDA model is
1 Corresponding
author. The final publication is available at http://dx.doi.org/10.1016/j.compositesb.2012.09.069
2 GraduateResearch Assistant.
3 Abaqus and Simulia are trademarks or registered trademarks of Dassault Systèmes or its subsidiaries in the United States

and/or other countries.

1
Composite Part B, 46:211–220, 2013. 2

applied for predicting other, independent results, and conclusions are drawn about the applicability of the
PDA model. This study focuses on E-Glass Epoxy Laminates.

2 Progressive Damage Analysis (PDA)

The Abaqus PDA model is a generalization of the approach proposed by Camanho and Davila [15]. It
requires linearly elastic behavior of the undamaged material and is used in combination with Hashin’s
damage initiation criteria [16].

2.1 Damage Initiation

In Abaqus [?] the onset of damage is detected by the Hashin and Rotem damage initiation criteria [16] in
terms of apparent (nominal, Cauchy) stresses σ, which is calculated by the FEA code. Damage initiation
refers to the onset of degradation at a material point. The criterion considers four different damage initiation
mechanisms, which are assumed to be uncoupled, as follows

Fiber tension (σ11 ≥ 0)

 2  2
σ11 σ12
Fft = +α (1)
F1t F6

Fiber compression (σ11 < 0)

 2
σ11
Ffc = (2)
F1c

Matrix tension and/or shear (σ22 ≥ 0)

 2  2
t σ22 σ12
Fm = + (3)
F2t F6

Matrix compression (σ22 < 0)

 2 " 2 #  2
c σ22 F2c σ22 σ12
Fm = + −1 + (4)
2F4 2F4 F2c F6

where σij are the components of the stress tensor; F1t , F1c , are the tensile and compressive strengths in
the fiber direction; F2t , F2c , are the tensile and compressive strengths in the matrix direction; F6 , F4 , are the
longitudinal and transverse shear strengths, and α determines the contribution of the shear stress to the fiber
tensile criterion. To obtain the model proposed by Hashin and Rotem [16] we set α = 0 and F4 = 1/2 F2C .
Furthermore, Ff t , Ff c , Fmt , Fmc , are indexes that indicate whether a damage initiation criterion in a damage
mode has been satisfied or not. Damage initiation occurs when any of the indexes exceeds 1.0.
The effect of damage is taken into account by reducing the values of stiffness coefficients [?, Section
21.3.1] as follows

σ=C:ε (5)

where : represents a tensor double contraction and the damaged stiffness is given by
 
(1 − df )E1 /∆ (1 − df )(1 − dm )ν21 E1 /∆ 0
C = (1 − df )(1 − dm )ν12 E2 /∆ (1 − dm )E2 /∆ 0  (6)
0 0 (1 − ds )G12
∆ = 1 − (1 − df )(1 − dm )ν12 ν21
ds = 1 − (1 − dtf )(1 − dcf )(1 − dtm )(1 − dcm ) (7)
Composite Part B, 46:211–220, 2013. 3

where σ is the apparent stress, ε is the strain, C is the damaged stiffness matrix, E1 , E2 , are the moduli in the
fiber direction and perpendicular to the fibers, respectively, G12 is the inplane shear modulus, ν12 , ν21 , are the
inplane and Poisson’s ratios, and dtf , dcf , dtm , dcm , ds , are damage variables for fiber, matrix, and shear damage
modes, in tension and compression, respectively. Note that the shear damage variable is not independent
but given by (7) in terms of the remaining damage variables.
The damage variables dtf , dcf , dtm , dcm , that characterize fiber and matrix damage in tension and compres-
sion correspond to the four damage initiation modes (1–4). At any time, each material point is either in
tension or compression. Therefore, the four damage variables reduce to two variables [?, Section 21.3.2] as
follows
 t
df if σ11 ≥ 0
df = (8)
dcf if σ11 < 0

and
dtm

if σ22 ≥ 0
dm = (9)
dcm if σ22 < 0

The six values of strength, F1t , F1c , F2t , F2c , F6 , F4 , are material properties that must be provided by the
user. Due to differences between testing and application conditions, as well as in-situ effects [1, Section 7.2.1],
the strength values measured by standard methods may not predict damage initiation accurately. Therefore,
in this work, the values of strength are found by correlating model results to laminate experimental data.
The same equations (1–4) are used as damage evolution criteria by introducing effective stresses (10) in
lieu of nominal stress. The relation between the effective stress σ e and the apparent stress σ is computed [17,
(8.64)] as

e = M −1 : σ
σ (10)

Abaqus PDA uses the model proposed by Matzenmiller et al. [18] to compute the degradation of the stiffness
matrix coefficients. Therefore, the damage effect tensor in Voigt notation is given as

(1 − df )−1
 
0 0
M −1 =  0 (1 − dm )−1 0  (11)
−1
0 0 (1 − ds )

Prior to damage initiation, the damage effect tensor M −1 is equal to the identity matrix, so σe = σ.
Once damage initiation and evolution has occurred for at least one mode, the damage effect tensor becomes
significant in the criteria for damage initiation of other modes. When df = 0 and dm = ds = 1, equation
(11) represents the degradation scheme of the ply discount method.

3 Damage Evolution
Once any of the damage initiation criteria (1–4) is satisfied, further loading will cause degradation of material
stiffness coefficients. The evolution of the damage variable employs the critical energy release rates Gci , which
are material properties, with i=ft,fc,mt,mc, corresponding to the four damage modes: fiber tension, fiber
compression, matrix tension, and matrix compression, respectively. Therefore, in addition to six values of
strength, four values of critical ERR must be provided, each one corresponding to the area of the triangle
OAC shown in Fig. 1. This contrasts with other models where both onset and evolution of damage are
predicted in terms of critical ERR only [19].
Normally, the constitutive model is expressed in terms of stress-strain equations. When the material
exhibits strain-softening behavior, as shown in Fig. 1, such formulation results in strong mesh dependency.
In Abaqus, to alleviate the mesh dependency in the PDA constitutive model, a characteristic length is
introduced into the formulation. For membranes and shells the characteristic length Lc is computed as the
square root of the area of the reference surface of the element. Using the characteristic length, the stress-
strain constitutive model is transformed into the stress-displacement constitutive model illustrated in Fig.
1.
Composite Part B, 46:211–220, 2013. 4

The evolution of each damage variable is governed by an equivalent displacement δ eq . In this way, each
damage mode is represented as a 1D stress-displacement problem (Fig. 1) even though the stress and strain
fields of the actual problem are 3D. The equivalent displacement for each mode is expressed in terms of
the components corresponding to the effective stress components used in the initiation criterion for each
damage mode. The equivalent displacement and stress for each of the four damage modes are defined as
follows [?, Section 21.3.3]

Fiber tension (σ11 ≥ 0)

q
ft
δeq = LC . hε11 i2 + αε212
ft hσ11 ihε11 i + ασ12 ε12
σeq = ft
(12)
δeq /Lc

Fiber compression (σ11 < 0)

fc
δeq = LC h−ε11 i
fc h−σ11 ih−ε11 i
σeq = fc
(13)
δeq /Lc

Matrix tension and shear (σ22 ≥ 0)

q
mt
δeq = LC hε11 i2 + ε212
mt hσ22 ihε22 i + σ12 ε12
σeq = mt /Lc
(14)
δeq

Matrix compression (σ22 < 0)

q
mt
δeq = LC h−ε22 i2 + ε212
mt h−σ22 ih−ε22 i + σ12 ε12
σeq = mc /Lc
(15)
δeq

where hi represents the Macaulay operator, which is defined as hηi = 21 (η + |η|) for every η ∈ <.
For each mode, the damage variable controls the reduction of the stiffness coefficients, and may assume
values between zero (undamaged state) and one (fully damage state). The damage variable for a particular
mode is derived using Fig. 1 and given by given by
C 0
δeq (δeq − δeq )
d= C − δ0 )
(16)
δeq (δeq eq

C
where δeq is the maximum value of δeq at point C in Fig. 1, for each mode.
Values of interlaminar critical energy release rates Gc reported in the literature are not directly applicable
as input to Abaqus PDA because the damage addressed by PDA is intralaminar, not interlaminar. Also,
the PDA model computes equivalent 1D energy release rates as the area under the curve on Fig. 1, not 3D
energy release rates, which are defined as
∂U
G=− (17)
∂A
where the strain energy U is a 3D function of ε, and A is the crack area. Furthermore, in PDA, there is no
identifiable crack but rather an equivalent reduction of stiffness. Therefore, the crack area A is not known
and the ERR cannot be computed as in (17), but in terms of equivalent 1D stress and displacement (14).
Composite Part B, 46:211–220, 2013. 5

4 Ply Discount
The ply discount method was used to compare to PDA results. The Hashin damage initiation criteria was
implemented in a user subroutine based on (1–4), using nominal stress.
For damage evolution, the ply discount method simply reduces to zero the stiffness of the lamina that
reaches the damage onset, except in the fiber direction which is left unchanged. In order to prevent conver-
gence problems, the secant stiffness (6) is reduced using df = 0, dm = ds = 0.999, abruptly as soon as the
Hashin damage initiation criteria (3) has been met.
Then, the stress is calculated with (5), where ε is total strain. In Abaqus, εi = εi−1 + ∆εi , where i is
the current step. If damage has not initiated, the damage initiation criteria is evaluated for the recently
calculated values of stress to decide whether or not the damage initiation criteria has been met. In ply
discount, damage evolution is sudden and abrupt, occurring simultaneously with damage initiation.

5 Experimental
Changes in the normalized longitudinal Young’s modulus and Poisson’s ratio with respect to the applied
strain, measured experimentally and reported in the literature, were used to compare with the Progressive
Damage Analysis and the Ply Discount methodology. The laminate stacking sequences reported by Varna
et al. in [20] and [21], shown in Table 1, were considered. The dimensions of the specimens are 20 mm wide
with a free length of 110 mm. The ply properties for the laminates are listed in Table 2.
All laminates were subjected to axial deformation εx , which coincides with ε11 in the 0◦ laminas. None of
the laminas in these laminates is subjected to fiber modes (1–2) or matrix compression (4). The 90◦ laminas
in laminates #1 and #5 are subjected to pure traction and no shear, so damage initiation is controlled
by the parameter F2t . Therefore, the laminate experimental data for laminate #1 (Fig. 2(a)) was chosen
to determine the insitu value of F2t , which is reported in Table 3. This value is 100% higher than the
UD strength reported in Table 2. When the insitu value is used to predict initiation for laminate #5, the
predicted first ply failure (FPF) strain is 22.4% higher than the observed experimental data (Fig. 3(a)). The
difference is due to insitu effect, because the damage initiation criteria used in Abaqus does not account for
insitu effect.

6 Methodology
All the laminates considered for the study are symmetric and balanced. Therefore a quarter of the specimen
was used for the analysis using symmetry b.c. and applying a uniform strain via imposed displacements on
one end of the specimen. A longitudinal displacement of 1.1 mm was applied to reach a strain of 2%. Abaqus
S8R elements were used for most of the study but the convergence study also considered S4R elements.
The value of apparent energy release rate Gcmt was obtained simultaneously with the value of F2t to
provide a best fit with the PDA model to the experimental modulus data for the laminate #1 (Fig. 2(a)).
Laminate #1 was chosen because only F2t and Gcmt are active. The experimental data and the fitted PDA
model results are shown in Fig. 2(a).
Since numerical minimization algorithms only find local minima, it is important to first find a good guess
for the adjustable parameters F2t and Gcmt . For this, 804 simulations were performed with random values of
the parameters in the interval 1.5 < Gcmt < 30.5 kJ/m2 and 21 < F2t < 141 M P a. The results are visualized
in Figs. 4 and 5. The error was calculated using the usual formula
v
uN
X E Abaqus exp !2
1u E 
Error = t  −  (18)
N i=1 E

e ε=εi E
e ε=εi

where E and E e now stand for the laminate longitudinal elastic modulus, damaged and intact respectively,
and N is the number of experimental data points. Nineteen experimental data points are reported by Varna
et al. [21].
Note that the shallow contour of error vs. Gcmt in Fig. 4 may explain the great dispersion of values
reported in the literature. In Fig. 4, a broad range of Gcmt (10 to 30 kJ/m2 in this case), seem to adjust more
Composite Part B, 46:211–220, 2013. 6

or less well the experimental degradation rate of modulus in Fig. 2(a). If the response measured (modulus)
is weakly sensitive to the parameter of interest (Gcmt ), the experimental values of the parameter will have
large scatter. Nevertheless, by adjusting simultaneously both parameters, F2t and Gcmt , the minimization
algorithm described next is able to converge to a global minimum.
From Figs. 4 and 5, it can be seen that a good choice for initial guess is Gcmt = 12 kJ/m2 and
F2t = 80 M P a. Next, a MATLAB R
script was executed to look for the minimum error (18), by repeatedly
executing Abaqus with parameters varying as per the Simplex method [22]. The converged values of F2t and
Gcmt are reported in Table 3.
According to (3), a combination of F2t and F6 determines the damage onset for off-axis plies. Since
the values of F2t and Gcmt are already set, it only remains to adjust the insitu value of F6 , which is done
by minimizing the error using the modulus data for laminate #8 because this is the laminate where the
cracking laminas experience the most shear. Since Abaqus PDA uses the Hashin damage initiation criterion,
equation (3) is active, F2t is fixed and only F6 can vary. A preliminary estimation of the error was obtained
on 112 simulations by using random numbers in the range 27.5 < F6 < 93 M P a. Varna et al. reported 13
experimental values [20, Fig. 13]. The error versus the longitudinal shear strength F6 is presented in Fig.
6. From this plot, the initial guess value for the search was chosen to be F6 = 49 M P a. Once again, the
MATLAB script was used to find the minimum error, at which point the converged value of F6 is found and
reported in Table 3.

7 Model Assessment
7.1 Comparison between PDA and Experimental Data
In this section we compare modulus and Poisson’s ratio predicted with PDA vs. experimental data for the
nine laminates listed in Table 1. Note that the modulus vs. strain data of laminate #1 was used previously
to find values for F2t and Gcmt , and that only the damage initiation in laminate #8 was relevant in finding
F6 . Since the remaining data has not been used to adjust the PDA parameters, it can be used for assessing
the quality of the predictions.
The longitudinal modulus and Poisson’s ratio for the nine laminates are compared with experimental
data in Figs. 2 and 7–13. Figure 2(a) merely states the fact that F2t and Gcmt were adjusted using the data
for this laminate. Figure 2(b) is a new result as it was not used to fit any parameters. It shows that the
laminate Poisson’s ratio is predicted well by the model.
As long as the cracking lamina is a 90◦ lamina (Figs. 3 and 7–9), PDA predictions are quite good, because
both the insitu F2t and Gcmt are adjusted using a cracking 90◦ lamina. Even when the cracking lamina is at
70◦ with respect to the load direction, the cracks propagate predominantly under mode I, so the predictions
of modulus is good (Fig. 10(a)).
However, cracking lamina under significant shear (Figs. 11–13) are not modeled well by PDA. This may
be the result of the shear damage variable ds not being an independent variable in the model. In fact, for the
experimental data available, all the remaining modes of damage are absent. Substituting dtf = dcf = dcm = 0
in (7) yields ds = dtm . That is, the reduction of shear modulus G12 represented by ds is modeled as being
equal to the reduction of transverse modulus E2 . This contrast with other models in the literature that use
independent variables for ds and dtm [23], etc. In fact, analytical computation of the reduction of stiffness after
solving the equations of elasticity for a cracked laminate in [19, (23)] for laminate #8 yields a constant ratio
ds /dtm = 0.69 from crack initiation to saturation crack density. In contrast, the PDA ratio is ds /dtm = 1.0.
Note that the data for laminate #8 was used to adjust F6 , an thus damage initiation is predicted
accurately in Fig. 12(a), but Gcmt was adjusted with laminate #1, and it cannot be readjusted with laminate
#8, so the damage evolution is not predicted accurately.

7.2 Insitu Strength

Since the values of F2t and F6 reported in Table 3 were calculated to match laminate data, the values
obtained are insitu values. To highlight the error incurred when using unidirectional (UD) lamina strength
values rather than insitu values, a comparison is presented on Table 4 between the First Ply Failure (FPF)
strain calculated with Abaqus and the FPF strain calculated using an online laminate analysis software
Composite Part B, 46:211–220, 2013. 7

(cadec-online.com) with and without insitu correction of the UD strength values. The UD strength values
(without insitu correction) were back calculated using [1, (7.42)] in terms of the insitu values reported in
Table 3 and assuming the transition thickness of the E-glass-epoxy material is 0.6 mm [24]. The Abaqus
results are found to be in good agreement with the FPF values calculated using insitu strength into the
Hashin damage initiation criterion implemented in the laminate analysis software [25]. However, as shown
in column 4 of Table 4, the FPF strain is grossly underestimated if the insitu correction is neglected.

7.3 Convergence
In this section we assess the sensitivity of the Abaqus PDA model to h- and p-refinement, realized by mesh
refinement and by changing the element type, respectively. Note that the specimen is subjected to a uniform
state of membrane strain, so h-refinement should not be necessary to converge on a stress field, unless the
constitutive model is mesh dependent.
However, the mesh convergence study reveals mesh dependency, as shown in Fig. 14. Before first ply
failure (FPF), the force-displacement response is independent of the number of elements used in the mesh.
After FPF, the force-displacement response diverge, revealing mesh dependency.
With n=140, asymptotic behavior at large strains is reached when the damage in the 90◦ lamina is fully
developed. At this stage the laminate modulus reduces to one third of the undamaged value, as shown in
Fig. 15. Also in Fig. 15, it can be seen that for an applied laminate strain εx = 2.2%, the 0◦ lamina fails
due to fiber tension, which is consistent with the fibers used in the laminates.
Modulus vs. strain for the [02 /904 ]S laminate #1 are reported in Fig. 15. It can be seen that the model
predictions are affected by the element type. Recall that the material parameters F2t , F6 , Gcmt , were adjusted
using quadratic elements (S8R). Reduced integration (S8R5) yields identical results, but when the element
is changed to linear (S4,S4R), the model predictions are significantly different, overestimating the modulus
reduction.

8 Conclusions
A novel methodology is proposed to determine the material parameters, namely insitu strength F2t , F6 , and
critical ERR Gcmt , using laminate experimental data. It is observed that PDA predictions are good for
mode I matrix cracking but deficient for mixed mode matrix cracking including shear (laminates #7, #8,
and #9). Also, the PDA model is not very sensitive to changes in the value of critical ERR, as shown by
the shallow relationship of model error vs. critical ERR. This may explain the large dispersion of values
reported in the litarature for the intra and interlaminar ERR. Compared to PDA, the ply discount method
severely overestimates the stiffness changes and leads to large errors in the prediction of energy dissipation
(toughness). However the asymptotic values of modulus reduction are reasonably accurate. Using PDA, all
FPF predictions are reasonably good as long as the ply thickness of the cracked lamina are similar (4 plies
in most of this study). However, a 20% insitu effect is shown when n=8 if the values of strength are not
readjusted for insitu effect taking into account the change in thickness from 4 to 8 plies.
When the fiber tension, fiber compression, and matrix compression are not active, which is normally the
case due to the high strength of the fibers and sudden brittle failure in compression, the PDA shear damage
is equal to the PDA transverse tensile damage. Such constraint is likely responsible for the poor prediction
shown when the cracking ply is subjected to shear. The present study used E-glass laminates to identify
the PDA parameters because the reduction of laminate modulus due to transverse matrix damage in carbon
fiber laminates is very small on account of the high modulus of the fibers. Therefore, it is unlikely that the
proposed methodology would be applicable to carbon fiber laminates since the parameter identification relies
on modulus vs. strain data. On the other hand, other models in the literature that are based on measurable
state variables, such as crack density, can be identified using crack density, which is significant regardless of
the fiber modulus.
Composite Part B, 46:211–220, 2013. 8

Acknowledgements
This work was the result of a semester-long project in the 2012 course MAE 646 Advanced Mechanics of
Composite Materials at West Virginia University, taught by the first author. The authors wish to express
their gratitude to their classmates for their comments and collaboration during the semester with special
mention and recognition to S. Koleti. Also, many thanks to Dr. A. Adumitroaie for introducing us to Abaqus
PDA, as well as to Rene Sprunger and SimuliaTM for their support of our educational programs.
Composite Part B, 46:211–220, 2013. 9

Figure 1: Equivalent stress versus equivalent displacement for a linearly softening material.

(a) Normalized modulus vs. applied strain. (b) Normalized Poisson’s ratio vs. applied strain.

Figure 2: Laminate #1 with LSS [02 /904 ]S .

Composite Part B, 46:211–220, 2013. 10

(a) Normalized modulus vs. applied strain. (b) Normalized Poisson’s ratio vs. applied strain.

Figure 3: Laminate #5 with LSS [0/908 /01/2 ]S .

Figure 4: Error vs. Gcmt .

Composite Part B, 46:211–220, 2013. 11

Figure 5: Error vs. F2t .

Figure 6: Error vs. F6 .

Composite Part B, 46:211–220, 2013. 12

(a) Normalized modulus vs. applied strain. (b) Normalized Poisson’s ratio vs. applied strain.

Figure 7: Laminate #2 with LSS [±15/904 ]S .

(a) Normalized modulus vs. applied strain. (b) Normalized Poisson’s ratio vs. applied strain.

Figure 8: Laminate #3 with LSS [±30/904 ]S .

Composite Part B, 46:211–220, 2013. 13

(a) Normalized modulus vs. applied strain. (b) Normalized Poisson’s ratio vs. applied strain.

Figure 9: Laminate #4 with LSS [±40/904 ]S .

(a) Normalized modulus vs. applied strain. (b) Normalized Poisson’s ratio vs. applied strain.

Figure 10: Laminate #6 with LSS [0/ ± 70/01/2 ]S .

Composite Part B, 46:211–220, 2013. 14

(a) Normalized modulus vs. applied strain. (b) Normalized Poisson’s ratio vs. applied strain.

Figure 11: Laminate #7 with LSS [0/ ± 55/01/2 ]S .

(a) Normalized modulus vs. applied strain. (b) Normalized Poisson’s ratio vs. applied strain.

Figure 12: Laminate #8 with LSS [0/ ± 404 /01/2 ]S .

Composite Part B, 46:211–220, 2013. 15

(a) Normalized modulus vs. applied strain. (b) Normalized Poisson’s ratio vs. applied strain.

Figure 13: Laminate #9 with LSS [0/ ± 25/01/2 ]S .

6000

5000
Force Fx [N]

4000

1 element
3000 2 elements
22 elements
140 elements
Elastic
2000
0.30 0.40 0.50 0.60 0.70 0.80
Displacement Ux [mm]

Figure 14: Force vs. displacement for laminate #8 using different elements. The elastic line goes through
the origin (not shown).
Composite Part B, 46:211–220, 2013. 16

Figure 15: Normalized modulus vs. applied strain for laminate #8 using different types of elements (E
ex =
68.22 GP a).
Composite Part B, 46:211–220, 2013. 17

Table 1: Laminates considered in this study.

LSS Source
1 [02 /904 ]S [21, Figs. 7]
2 [±15/904 ]S [21, Figs. 8]
3 [±30/904 ]S [21, Figs. 9]
4 [±40/904 ]S [21, Figs. 10]
5 [0/908 /01/2 ]S [20, Figs. 1,6,7]
6 [0/ ± 70/01/2 ]S [20, Figs. 2,9,10]
7 [0/ ± 55/01/2 ]S [20, Figs. 3,11,12]
8 [0/ ± 40/01/2 ]S [20, Figs. 13,14]
9 [0/ ± 25/01/2 ]S [20, Figs. 15,16]

Table 2: Unidirectional ply properties for glass/epoxy Fiberite HyE 9082Af. See Table 3 for insitu strengths.

Property Units Value Source

E1 [GPa] 44.7 [20]
E2 [GPa] 12.7 [20]
G12 [GPa] 5.8 [20]
G23 [GPa] 4.5 [19]
ν12 [GPa] 0.297 [20]
α1 [µε/K] 8.42 [26]
α2 [µε/K] 18.42 [20]
∆T [K] -99 [20]
F1t [MPa] 1020 [17]
F2t [MPa] 40 [17]
F1c [MPa] 620 [17]
F2c [MPa] 140 [17]
F6 [MPa] 60 [17]
ply thickness [mm] 0.144 [21]

Tables
References
[1] E. J. Barbero. Introduction to Composite Materials Design–Second Edition. CRC Press, Boca Raton,
FL, 2nd edition, 2010.
[2] R. J. Nuismer and S. C. Tan. Constitutive relations of a cracked composite lamina. Journal of Composite
Materials, 22:306–321, 1988.
[3] S. C. Tan and R. J. Nuismer. A theory for progressive matrix cracking in composite laminates. Journal
of Composite Materials, 23:1029–1047, 1989.
[4] S. Li, S. R. Reid, and P. D. Soden. A continuum damage model for transverse matrix cracking in
laminated fibre-reinforced composites. Philosophical Transactions of the Royal Society London, Series
A (Mathematical, Physical and Engineering Sciences), 356:2379–2412, 1998.

Table 3: Insitu strengths and Apparent Energy Release Rate.

Property Units Value

F2t [MPa] 80.8831
F6 [MPa] 48.5725
Gcmt [kJ/m2 ] 11.5002
Composite Part B, 46:211–220, 2013. 18

Table 4: Comparison of FPF laminate strain εx using Abaqus with insitu values and cadec-online.com with
and without insitu correction of strength.

LSS Abaqus cadec-online.com

with insitu with insitu without insitu
1 [02 /904]S 0.6420 0.6420 0.4053
2 [±15/904 ]S 0.6622 0.6452 0.4073
3 [±30/904 ]S 0.6622 0.6511 0.4111
4 [±40/904 ]S 0.6420 0.6387 0.4121
5 [0/908 /01/2 ]S 0.6420 0.6389 0.4034
6 [0/ ± 704 /01/2 ]S 0.6219 0.6094 0.3972
7 [0/ ± 554 /01/2 ]S 0.5817 0.5918 0.3837
8 [0/ ± 404 /01/2 ]S 0.5616 0.5514 0.3879
9 [0/ ± 254 /01/2 ]S 0.7427 0.7077 0.5069

[5] E. Adolfsson and P. Gudmundson. Matrix crack initiation and progression in composite laminates
subjected to bending and extension. International Journal of Solids and Structures, 36:3131–3169,
1999.
[6] J. A. Nairn. Polymer Matrix Composites, volume 2 of Comprehensive Composite Materials, chapter
Matrix Microcracking in Composites, pages 403–432. Elsevier Science, 2000.
[7] J. M Berthelot. Transverse cracking and delamination in cross-ply glass-fiber and carbon-fiber reinforced
plastic laminates: static and fatigue loading. Applied Mechanics Review, 56:111–147, 2003.
[8] J. A. Nairn. Finite Fracture Mechanics of Matrix Microcracking in Composites, pages 207–212. Appli-
cation of Fracture Mechanics to Polymers, Adhesives and Composites. Elsevier, 2004.
[9] S. H. Lim and S. Li. Energy release rates for transverse cracking and delaminations induced by transverse
cracks in laminated composites. Composites Part A, 36(11):1467–1476, 2005.
[10] D. T. G. Katerelos, J. Varna, and C. Galiotis. Energy criterion for modeling damage evolution in
cross-ply composite laminates. Composites Science and Technology, 68:2318–24, 2008.
[11] D. T. G. Katerelos, M. Kashtalyan, C. Soutis, and C. Galiotis. Matrix cracking in polymeric composites
laminates: Modeling and experiments. Composites Science and Technology, 68:2310–17, 2008.
[12] J. A. Mayugo, P. P. Camanho, P. Maimi, and C. G. Dávila. Analytical modeling of transverse matrix
cracking of [±θ/90n ] composite laminates under multiaxial loading. Mechanics of Advanced Materials
and Structures, 17:237–45, 2010.
[13] GENOA Virtual Testing & Analysis Software, Alpha STAR Corporation, http://www.ascgenoa.com.
[14] Helius:MCT Composite Materials Analysis Software, Firehole Composites, http://www.firehole.com/
products/mct/.
[15] P. Camanho and C. Davila. Mixed-mode decohesion finite elements for the simulation of delamination
in composite materials. NASA/TM-2002-211737, pages 1–37, 2002.
[16] Z. Hashin and A. Rotem. A fatigue failure criterion for fiber-reinforced materials. Journal of Composite
Materials, 7:448–464, 1973.
[17] E. J. Barbero. Finite Element Analysis of Composite Materials. Taylor & Francis, 2007.
[18] A. Matzenmiller, J. Lubliner, and R. Taylor. A constitutive model for anisotropic damage in fiber-
composites. Mechanics of Materials, 20:125–152, 1995.
[19] E. Barbero and D. Cortes. A mechanistic model for transverse damage initiation, evolution, and stiffness
reduction in laminated composites. Composites Part B, 41:124–132, 2010.
Composite Part B, 46:211–220, 2013. 19

[20] J. Varna, R. Joffe, N. Akshantala, and R. Talreja. Damage in composite laminates with off-axis plies.
Composites Science and Technology, 59:2139–2147, 1999.
[21] J. Varna, R. Joffe, and R. Talreja. A synergistic damage mechanics analysis of transverse cracking in
[±θ/904 ]s laminates. Composites Science and Technology, 61:657–665, 2001.

[22] J. Lagarias, J. Reeds, M. Wright, and P. Wright. Convergence properties of the nelder-mead simplex
method in low dimensions. SIAM Journal of Optimization, 9:112–147, 1998.
[23] E. J. Barbero and L. De Vivo. Constitutive model for elastic damage in fiber-reinforced pmc laminae.
International Journal of Damage Mechanics, 10(1):73–93, 2001.
[24] A. S. D. Wang. Fracture mechanics of sublaminate cracks in composite materials. Compos. Tech.
Review, 6:45–62, 1984.
[25] Computer Aided Design Environment for Composites (CADEC) http://en.cadec-online.com.
[26] E. J. Barbero, G. Sgambitterra, A. Adumitroaie, and X. Martinez. A discrete constitutive model for
transverse and shear damage of symmetric laminates with arbitrary stacking sequence. Composite
Structures, 93:1021–1030, 2011.